Hi there, I have been attempting to submit an audio file for transcription, and as far as I can tell, I’ve formatted the HTTP POST request correctly, but the server is responding with an error: “you must provide a model parameter”.
Here is the HTTP POST request:
[header] POST /v1/audio/transcriptions HTTP/2 Host: api.openai.com User-Agent: curl/8.4.0 Accept: */* Authorization: Bearer [redacted] Content-Length: 654410 Content-Type: multipart/form-data; boundary=F8DD6D1A-530A-464B-921D-1E30EC63F8A5
Doh. Figured it out. Several typos: after, name=“file”, I used a comma instead of a semi-colon; after filename, I used a colon, instead of an equals sign.
Just in case anybody else stumbles upon this topic because the server cannot parse your multipart/form request:
The OpenAI API currently cannot handle boundaries that include dashes (-). If you construct your request manually or the web client you use inserts dashes as part of the boundary, you might need to redefine these boundaries. I think this is a bug in the API since RFC1341 explicitly permits this: